7th International Symposium on Geotechnical Aspects of Underground Construction in Soft Ground; 16-18 May 2011, Rome, Italy

Design of jet-grouting for tunnel waterproofing

Marcos Arroyo & Antonio Gens
Departament of Geotechnical Engineering and Geosciences, UPC, Barcelona, Spain
Paolo Croce & Giuseppe Modoni
University of Cassino, Italy

ABSTRACT: A case history in Barcelona is described where a tunnel was excavated by traditional methods
below an active railway line through a formation with lenses of water-bearing granular material. To avoid the
possibility of sudden collapses a massive jet-grout treatment was applied. The treatment took several forms.
Subvertical double and triple-fluid injection was applied whenever possible. Sub-horizontal monofluid cano-
pies and slabs executed from within the tunnel were however required in zones where no vertical access was
possible. This communication focuses on the later type of treatments and gives an overview of the design
tools that were applied. These comprised the execution of several large trial fields and the systematic applica-
tion of a probabilistic framework for design.

1 INTRODUCTION able for all improvement purposes (strengthening,

It is a common paradox that surface transport infra-
structures are usually more needed where there is
less space available for them. Underground devel-
opments are then inevitable, be that by means of tun-
nels and excavations, or through refurbishment and
adaptation of existing infrastructures, requiring new
foundations and/or extensions. Wherever urban in-
frastructure construction activities take place there is
a large likelihood that one form or another of ground
improvement would be required. Ground improve-
ment allows construction to proceed where other-
wise it would be impossible, because some relevant
required soil property (strength, stiffness, permeabil-
ity) is missing.
Jet grouting is a technique where a high-pressure
injection of mortar, with or without other accompa-
nying fluids (water, air), impacts the ground in a
borehole. In most cases the original ground is thus
eroded, mixed with the mortar and, in fluid form,
partly evacuated to the surface (resulting on what is
called “spoil”). The remaining soil-cement mixture
sets “in situ”, resulting on a stiffer, stronger, more
impermeable and less ductile material than the origi-
nal soil. The injection equipment is displaced along
the borehole, thus creating a body of treated soil of
columnar shape. Several such injections are com-
bined to create the desired shape of treated soil:
slabs, arches and walls are common examples.
The basic reasons for jet-grout success are clear:
of all the means of ground injection, jet is not only
the fastest procedure, but is also the only one suit-
stiffening, impermeabilization). In tunnelling opera-
tions jet grouting has the added attractive of access
versatility: treatments can be executed from within
the tunnel, from the surface of the ground or, even-
tually, from a side shaft.
This paper describes a case history in Barcelona
where a variety of jet-grout treatments were per-
formed to help drive a tunnel through water-bearing
sediments under an active railway line. After giving
the geotechnical background of the problem as well
as a brief account of the construction procedures ap-
plied, the paper focuses on the design tools em-
ployed. Several results from large-scale “in situ”
tests are described; afterwards an example of a prob-
abilistic approach employed in design is outlined.
Finally a brief summary of the observed treatment
outcomes is also given.
2 PROJECT DESCRIPTION
2.1 Location
The new High Speed Railway Link between Madrid
and Barcelona was open to traffic in February 2007.
The Southern entry of the new line into the city of
Barcelona follows the trace of the historical railway
entrance, heading towards Sants railway station
(Figure 1). This railway entrance crosses the city of
Hospitalet de Llobregat, just south of Barcelona.
Hospitalet is currently involved in a major urban re-
design operation in which the suppression of the
railway barrier plays a major role. For this reason
the design of the new high speed link located it be-
low ground well before its arrival at Sants.
In plan, the new link follows closely the current
tracks, so much that the final 2 km stretch (marked
IV in Figure 1), had to be excavated just below
them. The railway line south of Sants is the main
medium and long-distance line out of Barcelona,
doubling also as a very busy commuter line. There-
fore only limited railway traffic restrictions were
possible and the new entry had to be built below the
still active old lines.

2.2 Ground conditions

The tunnel is located on the alluvial plain of Barce-
lona. The western limit of this geomorphological
unit is given by the Collserola mountain range,
whereas the southern and northern limits are given
by the deltaic systems of two river mouths (Llobre-
gat and Besos, respectively). The alluvial plain is
formed by Quaternary deposits overlying a Tertiary
substratum. The main unit of the Quaternary depos-
its in the tunnel area is a brown red clay that in-
cludes some carbonated levels and thin sand layers
(QPA in Figure 2). This unit is crossed by a network
of paleochannels, fossilized remains of a network
draining the nearby Collserola range into the Medi-
terranean. These paleochannels are formed by sands
and gravels with a variable clay matrix (QPAr).
Younger, recently active brook deposits are also
crossed by the tunnel trace (QR). The Tertiary depos-
its, mostly from Pliocene and Miocene age, are
formed by some ochre sands and clays (PA) and
blue-grey marine marly clays (PM). The water table
in the area is generally located between 10 and 15 m
depth.
Figure 1. Plan view of the High Speed Railway Link south en-
trance to Barcelona.
Figure 2. Geotechnical profile alongside the central part of the
tunnel. PM stiff tertiary marl; QPA medium stiff clay-dominated
quaternary deposits; QPAr sand-dominated quaternary deposits;
QR gravel-dominated quaternary deposits; QANT made ground.
Spacing amongst vertical reference lines 20 m; spacing
amongst horizontal reference lines 2 m.
2.3 General tunneling concept
The main tunnel section had nearly 110 m2. The lon-
gitudinal profile was established to pass below sev-
eral underground road passages and a tube line that
marked its lowest point. The tunnel cover varied be-
tween a minimum of 6 m and a maximum of 25 m.
In the situation just described the main construction
priority was to avoid any risk of sudden ground col-
lapse, however minor. On the other hand, and thanks
to an active ballast maintenance program, slowly in-
duced excavation settlement could be easily com-
pensated.
EPB tunnelling is generally the favoured urban
tunnelling choice when soft soils are present on a
tunnel trace. However, for relatively short tunnels or
when access shafts for an EPB are not easily located,
traditionally mined excavation might offer an inter-
esting and robust alternative. These two circum-
stances were present at the tunnel here described. A
partitioned section procedure, known as the tradi-
tional Madrid or Belgian Method, was then selected
as the basic excavation procedure for the tunnel. It is
worth mentioning, however, that this method is rela-
tively slow, with median advance rates of 30
m/month per excavation front (Sacyr 2008). To
comply with a tight construction schedule this re-
quired the execution of 2 intermediate deep access
ramps and 3 further deep shafts to open intermediate
excavation fronts.
The first 800 m of the tunnel were located either
above the water table or within the impermeable,
homogeneous PM layer. There, the Belgian method,
with occasional forepoling help, was successfully
applied (for details, see Deu et al. 2007). The last
600 m of the tunnel were again located mostly above
the water table, and a similar construction procedure
was applied.
 7th International Symposium on Geotechnical Aspects of Underground Construction in Soft Ground; 16-18 May 2011, Rome, Italy

However, in the approximately 400 m corre- chamber by covering the full section perimeter and
sponding to the central and deepest part of the tun- closing the full tunnel face ahead of the excavation
nel, the presence of erratic channels of quaternary using sub-horizontal columns. From within the ex-
deposits with a large granular fraction (QPAr, QR) cavated heading followed a full treatment of the
was coincident with a water table well above the bench with subvertical columns. Executing the in-
tunnel crown (Figure 2). There were also close jection from within the tunnel itself only allowed for
precedents showing that running ground conditions single-fluid injection and relatively small diameter
at the tunnel face were a distinct possibility. This columns (around 0.5 to 1 m, see below for details).
was a major cause for concern, and thus there was
some consensus that even the small partitioned and
braced excavation fronts allowed by the Belgian
method did not offer enough guarantee against that
kind of failure. Therefore in that part of the tunnel a
different construction procedure, relying on a sys-
tematic use of jet-grout, was envisaged.
Table 1 Geotechnical parameters; average values in parenthe-
ses
Material Fines w NSPT Su E
% % kPa MPa
QANT 80-9 (50) (15.1) (5) - 7.5
QR 86-14 (48) - 22.5 (a) (b)
(16.1)
(43) Figure 3. Typical treatment configurations where surface ac-
QPa 99-50 (24) (100) 22.5 cess was available.
(18.4)
(80)
QPar 80-5 (30) (14.5) (31) - 40
PM 23-99 (34) (150) 42
(20.8) sub-horizontal columns

(95)
I phase sub-horizontal columns
I phase
sub-vertical columns

II phase

2.4 Jet-grouting treatments sub-vertical columns

Jet-grout was selected as the soil improvement tool

that was best suited to avoid the possibility of
ground flows towards the excavation. The main
function of the treatment was then one of imperme- sub-horizontal columns
I phase
abilization around the tunnel; the natural soil had
more than adequate resistance and stiffness to span II phase
the tunnel section without excessive deformations
(Table 1).
sub-vertical columns
Two types of treatments were applied. In loca-
tions where access above the treated zone was pos-
sible, vertical or subvertical columns were
Figure 4. Typical configuration for treatment from within the
employed. For this type of treatments powerful two- tunnel (transverse and longitudinal sections).
and three-fluid means of jet injection could be
applied, since the spoil might be evacuated from the These zones executed from within the tunnel
surface and the geometry of a newly executed were the most critical, both because the treatment
column was not unstable. The typical design column execution was particularly risky and because the re-
diameter for the vertical columns was 2.5 m. The quired number of columns made the procedure very
layout included full face sections, where the whole slow. A number of special measures were then
tunnel face was covered by the treatment (Fig. 3a) taken to guarantee a successful outcome, both at the
and sections where the columns where only design stage and at the execution stage. While this
surrounding the tunnel (Fig. 3b). The purpose of the paper focuses on design, it is worth mentioning,
full face sections was to create longitudinally amongst the execution-related aspects, the system-
isolated excavation zones, so as to minimize the atic application of a manually controlled blow-out
extension
However,of hypothetic
there werefailures.
some zones within the preventer (Guatteri et al. 2009), a tool that helped to
treated tunnel section where surface access above avoid instabilities with the sub-horizontal injections.
the tunnel was impossible. In these zones the exca-
vation was partitioned between heading and bench
(Figure 4). Treatment of the heading created a soil
3 JET-GROUT DESIGN TOOLS
3.1 General approach
The development of jet-grout has been technologi-
cally driven, with the relevant science lagging be-
hind. As a result the design, implementation and
control of jet-grout treatments has developed on a
heuristic basis, lead by practitioners and real-life ex-
perience applications. Although this approach has
been successful in most cases, a number of high-
profile construction problems can be traced to inef-
fective jet-grout treatments, for instance the Souter-
rain tram tunnel in The Hague (NL) (van Tol 2004),
the Les Cretes tunnel of the Aosta highway in Italy
(Croce et al. 2004) and the Kaoshiung Mass Rapid
Transport tunnel, in Taiwan (Ishihara 2008).
The effectiveness of jet grouting treatments is
dictated by the interaction between technological
factors and natural soil properties, the role of which
cannot yet be predicted with sufficient accuracy but
needs to be evaluated by means of experimental in-
vestigations. To prove the efficiency of treatments,
execution of preliminary field trials is in fact re-
quired by current standards (e.g. ENV 12716 1997).
In design cases where the treatment should provide
watertightness, continuity of treatment is the most
fundamental aspect. Field trials should then focus on
the cross sectional dimensions and on the required
spacing of contiguous columns supposed to overlap.
Concerning the former point it is worth remind-
ing that, for a given set of injection parameters, col-
umn diameters are strongly influenced by natural
soil properties and that inhomogeneous subsoil con-
ditions turn into irregularities of column shape and
discontinuity of the waterproofing barriers (e.g.
Croce & Modoni 2005). With regard to the second
aspect, even if a direction is specified for columns
axes, deviations are possible depending on how ac-
curately the position and inclination of injection
tools is controlled. Generally, even when low toler-
ances are prescribed, unpredictable misalignments
occur, particularly for longer columns, due to the
self weight of perforation bars or to inaccurate driv-
ing of the supporting arms.
Bearing in mind these goals, different field trials
were specifically devised in the presented work,
some to find relations between column diameters
and treatment parameters, others to quantify the ef-
fects of imperfect drilling operations.
All the information collected in the trials was
then fed into probabilistic design models to explic-
itly take into account the inherent variability of the
method and deal economically with the uncertainty
of the treatment outcomes.
Figure 5. Plan view: vertical column trials.
3.2 Vertical column trials
A first field trial was located within one of the in-
termediate deep access ramps to the tunnel. The
ramp itself was roughly at the middle of the stretch
were the treatment was required, however, specific
stratigraphy at the ramp site was dominated by the
more clayey quaternary levels (QPa).
Two parallel rows of ten vertical columns each
(identified as A and B in Figure 5) were injected for
a length of about 15 m starting at 2 m below the sur-
face. Four columns of each group (A1, A8, A9, A10
and B1, B8, B9, B10) were positioned with spacing
larger than the expected diameter (0.5 m) to provide
data in diameter variability and material properties.
The other six were intended to overlap being in-
jected with variable axes span (0.30 m for A2, A3,
A4 and B2, B3, B4 columns, 0.45 m for A5, A6, A7
and B5, B6, B7).
Table 2. Injection parameters of vertical trial columns.
v
Column
(cm/min p (bar) V0 (m/s) ρ (t/m3) W/C
type
)
A 37 350 200 1.59 1/1.2
B 35 400 212 1.52 1/1
All columns were executed with a single-fluid
system, like the one to be applied later within the
tunnel. Columns A and B shared some injection pa-
rameters like nozzle diameter (3 mm) and rotation
velocity (12 rpm). However, they did differ in sev-
eral parameters (Table 2): withdrawal speed, v; in-
jection pressure, p (and, consequently, nozzle veloc-
 7th International Symposium on Geotechnical Aspects of Underground Construction in Soft Ground; 16-18 May 2011, Rome, Italy
ity, V0); water/cement ratio, W/C (and, conse-
quently, grout density, ρ).
The treatment parameters for both columns had
been initially proposed by the contractor, whose ex-
perience-based average diameter estimate for both
types of treatment was equal to 50 cm.
It seemed clear, however, that other predictive
approaches to estimate column diameter were also
worth trying. For instance, the treatment input en-
ergy per unit length could be computed following
Croce & Flora (2000) as
8  V03
En 
 vd2
In the case of the vertical trial columns type A
columns had En = 7.36 MJ/m, whereas type B col-
umns had En = 9.06 MJ/m. It was then expected that
type B columns would result in higher column di-
ameters and, because of their higher W/C ratio, in
smaller strengths.
Several treatment outcomes were measured in the
field trial. On the one hand, continuous rotary coring
of four columns provided samples at every meter on
which to measure strength (unconfined compression,
qu), stiffness (on-sample deformation measurements,
E50) and density of the treated material. On the other
hand the top four meters of all the trial columns
were examined while excavating the station access
ramp and cross sectional dimensions were taken by
measuring the length of cemented soils samples
cored by horizontal drilling. A summary of the re-
sults obtained through these measurements is in-
cluded in Table 3.
Table 3. Measured treatment outcomes on vertical trial col-
umns
Column COV qu COV
D (m) E50/ qu
type (D) (MPa) (qu)
A 0.38 0.14 11.6 0.26
B 0.48 0.12 7.9 0.38
The average diameter of columns B was higher
than the average diameter of columns A, in agree-
ment with the computed energy input (1). The
strength of columns B was less than that of columns
A, again as expected from the W/C relation.
Strength variability was higher than geometric vari-
ability. The latter was also easily quantified by
means of the distributions of column diameters
(Figure 6).
It was made clear that, while detrimental for the
mechanical properties of the treated soils, a larger
amount of water in the injected mix enhanced the
erosive action of the jet and resulted in larger col-
umn diameters. For the case here contemplated the
reduction in strength was immaterial, whereas the
increase in diameter was clearly beneficial.
A second consequence of this first field trial was
to show the benefits that might follow from a ra-
tional approach to diameter prediction. A more
elaborated predictive formula, following from Mo-
doni et al. (2006), was then proposed and validated
with a later round of single-column field trials. It is
not possible here to give a detailed account of this
latter round of single-column trials. However, it is
worth mentioning that the sub-horizontal columns
that were finally adopted as a basis for the treatment
execution were injected with larger nozzle diameter
(3.5 millimeters), with higher water-cement ratio
(1.5/1), and with a lower lifting speed (20 cm/min)
(1)
than those originally proposed.
0.4
Columns A-B (C.o.V.= 0.13)
.
Columns inside tunnel (C.o.V.= 0.17)
Frequency distribution
0.3
0.2
0.1
0
0.6 0.8 1 1.2 1.4
D / D
Figure 6. Distributions of column diameters measured at the
vertical field trials.
3.3 Full-scale heading trial
A second, more ambitious, field trial took place to
test the execution of the treatment conceived for the
heading section of the tunnel. An almost full-scale
(80%) heading section was built with a similar ge-
ometry and procedure as that intended in the tunnel
(Figure 4) from a dedicated 15 m deep excavation.
573
576
Figure 7. Profile of settlements at ground level induced by exe-
cution of treatments and extraction of water from the tunnel
chamber.
For reasons of space availability the trial took
place at some 8 km to the south of the tunnel loca-
tion. This meant that instead of on the Barcelona al-
luvial plain the trial was fully within the Llobregat
delta plain. The geotechnics of the Llobregat delta
are rather different from that of the alluvial plain,
with a medium dense sand layer overlaying a soft
sandy and clayey silt and water table close to the
surface (e.g. Gens et al. 2011). The trial heading sec-
tion was mostly performed within the silt layers and
had 9 m soil cover above the crown, mostly in sands.
The geotechnical conditions for the jet-grout execu-
tion at the trial site were then clearly unfavorable
when compared to those prevailing at the tunnel site.
However, it was felt execution under such condi-
tions will clearly prove the potential and/or show the
limitations of the proposed construction technique.
The tunnel contour canopy consisted of two con-
centric rows of 151 diverging columns, 12 m long
and with 0.50 m expected diameters, injected after
initial perforation of a previously created thick jet
grouting supporting wall. A 3 m thick plug was cre-
ated at the deeper end of the tunnel, made of 174
columns of 0.80 m diameter, located on six concen-
tric circles with 0.50 increasingly larger radii.
A net of vibrating wire piezometers and settle-
ment plates was installed above and around the trial
tunnel for monitoring purposes. Treatment execution
parameters were continuously recorded and, in sev-
eral instances, measurements of column axis inclina-
tion were also performed.
S01-16E S01-16D S01-15E
S00-16E S01-14D
-0.05
0.00
0.05
z (m)
0.10 designed direction - x
y 
0.15

0.20 true direction
z
0.25
0.00 0.05 0.10 0.15 0.20 0.25
y (m)
Figure 8. Deviation of column axes from their theoretical posi-
tion (deviation projected on a plane orthogonal to column
axis).
Here only a few results from this trial will be
commented; some results can also be found in Guat-
tieri et al. (2007). During jet injection asymmetrical
settlements and heave were recorded at ground level
(about 20 millimetres maximum settlement on one
side and 10 mm heave on the other side, Figure 7).
These movements were likely related to a pressure
build-up induced by injection. However, water pres-
sure measurements were almost constant in all the
piezometers.
While this outcome was hardly welcome, it was
not very discouraging. On the one hand, settlement
issues were of limited concern (although the possi-
bility of brusque jet-induced motion limited the tem-
poral window for treatment in certain cases). On the
other hand, the construction of the railway tunnel
had to be performed at considerably larger depths
below a thicker soil cover. Even in the absence of a
detailed model, it was clear that the movements that
were induced at the trial were well above those to be
expected on site.
After conclusion of treatments, a test on the wa-
terproofing capacity of the jet grouted soil was per-
formed, by inserting a number of pipes at the front
of the tunnel and by extracting, with a vacuum
pump, the water present in the inner chamber.
Pumping lasted two days and the amount of water
extracted was well below the estimated soil water
content (less than 5%). The poor response of the
pipes was likely due to clogging of the pipe protec-
tive filters by silt particles. During this test, a sudden
drop of water head was measured in a piezometer
located on the right side of the tunnel. Moreover,
this was immediately followed by a funnel shaped
collapse that emerged at the ground surface in a po-
sition close to the contact of the canopy and the ex-
treme plug of the tunnel (see settlement profile of
Fig.6).
Explanation for this result can be found in the
consolidation of soil inside the tunnel chamber acti-
vated by the extraction of water and by the lack of
bending capacity of jet grouting. It is worth men-
tioning that jet grouting columns were not reinforced
and that collapse occurred at the attachment of col-
umns to the end plug, i.e. where the flexural moment
reaches its maximum values. The amount of con-
solidation settlement to be expected at the tunnel site
was much less and therefore it was concluded that
passive (i.e. without pumping) dewatering of the soil
chamber could be applied safely at the tunnel.
Also particularly interesting are the measure-
ments of column axes inclination, which are rather
infrequently performed in jet grouting treatments
0.30
within tunnels. These were measured using one in-
clinometer mounted on the injection mast (Jean Lutz
2006). The two components Δy and Δz of deviation
from the theoretical axis direction, plotted for five
different columns in Figure 8, show that there is a
common systematic downward trend. Deviations in-
crease significantly with column length, and their
absolute values are non-negligible.
The deviations could be explained as a result of
the bending of perforation bars due to their self
weight (which, for the measured columns, was also
increased by the inclinometer). Partial compensation
can be provided by an initial upward inclination of
the perforation axis, as shown by column S01-15E.
 7th International Symposium on Geotechnical Aspects of Underground Construction in Soft Ground; 16-18 May 2011, Rome, Italy

0.35
analyzed during the project by developing probabil-
istic methods specifically customized for each struc-
 =0.68°
0.3 ture. An analysis developed for the perimetral cano-
s=0.17 °
pies of a tunnel heading (Figure 4) is here reported
0.25 to illustrate these calculations.
The probabilistic analysis of this structure was
frequency

0.2
performed selecting two sections, at the extremes of
0.15 the overlap section between two consecutive cano-
pies (see Figure 11a). Based on the field trial results
0.1 as well as on later quality assurance measurements,
0.05
a probabilistic distribution was assigned to column
diameters and column axes deviation. In particular, a
0 Gaussian distribution has been considered for col-
0 0.5 1 1.5 umn diameters with a mean value equal to 0.75 m
 and a coefficient of variation equal to 0.17. This
Figure 9. Frequency distribution of the inclination deviation value of average diameter has been chosen as equal
angles Δβ. to the minimum value estimated with the adopted set
of injection parameters for sandy soils, considering
these latter as the most critical in case of piping.
0.4 Random divergence of column axes from their
 = 29.45° designed position has been modelled by considering
0.35 angles Δα and Δβ respectively distributed with uni-
s = 23.94°
0.3 form and Gaussian laws. For the latter a nil average
value and a standard deviation of 0.68° were as-
0.25
frequency

signed.
0.2 A Monte Carlo simulation technique was applied,
0.15
generating one thousand cross sections consistently
with the adopted probabilistic distributions (one ex-
0.1 ample of simulation result is reported in Figure 11b).
0.05 The minimum canopy thickness was then calculated
in each case to obtain a statistical sample of this
0 simulation outcome. The cumulative frequency dis-
0 20 40 60 80 tribution of this variable (Figure 11c) represents the
 discontinuity hazard of this particular structure.
Figure 10. Frequency distribution of the azimuth deviation an-
gles Δα.
8
Again, and despite is large magnitude, this sys- A B

tematic deviation was not a major cause for concern.

Cross section A
Cross section B

4
The continuity of a jet grouting canopy is not par-
ticularly sensitive to systematic deviations. On the 0

other hand a random scattering of column directions 0 4 8

would be very detrimental for the same objective. -4

The deviation data were analyzed to quantify the (a) (b)

random scatter in direction. The average direction of 1

columns was calculated and the two angles (azimuth

Cumulated frequency

0.8 section B
Δα and divergence Δβ) expressing the deviation of 0.6 section A

columns axes from this mean trend were evaluated. 0.4

The plot in Figure 9 shows that a bell shaped fre- 0.2
quency distribution can be clearly inferred for the 0
divergence angle Δβ, while Figure 10 shows that the 0 0.1 0.2 0.3 0.4 0.5
distribution of azimuth angle Δα is approximately Minimum canopy thickness (m)

uniform. (c)

Figure 11. Probabilistic analysis of the continuity of tunnel

canopy (a. longitudinal section; b. example of cross section
simulation; c. cumulated frequency distributions of the mini-
3.4 Probabilistic approach mum thickness of the canopy).

The continuity of different jet grouting structures

(canopies, plugs and vertical walls) was routinely
4 JET GROUT PERFORMANCE
The field trials and probabilistic methods described
above were useful to generate confidence on the
construction team about the issue of one construc-
tion operation that was initially perceived as ex-
tremely risky. About 100 m of the main tunnel and
access galleries were thus treated in full section or
just at the bench from within the tunnel. The ad-
vance was performed without any incident.
However, and despite the optimization allowed
by the probabilistic design, the complexity of the
subterranean treatments still resulted in very slow
excavation procedure, with median advance rates
below 1m/day. An effort was then made to allow
even some partial surface access above most of the
treated tunnel length and therefore vertically exe-
cuted treatments were applied in most of the treated
section. Lacking time to perform the same extensive
trials that preceded the subhorizontal treatments, the
uncertainty on the jet grouting outcomes for vertical
treatments was higher, particularly for column di-
ameter. Hence ancillary pumping to lower the water
pressure on the treatment walls was introduced as an
auxiliary measure to increase the safety of excava-
tion. The combination of partial pumping and verti-
cal jet-grout chambers was successful in that the
tunnel was excavated without incidents at a median
advance rate two to three times higher than that al-
lowed by subhorizontal treatment.
5 CONCLUSION
The reported case history has proved that the design
of jet grout structures can largely benefit from a ra-
tional approach that includes experimentally based
uncertainty within a probabilistic framework. The
measurements taken and observation made at the
field trials in this project contain useful information
for future designs in similar conditions.
6 ACKNOWLEDGEMENTS
The authors acknowledge the opportunity given by
ADIF to participate in the studies described in this
work. The contribution of the Spanish Ministry of
Science through research grant BIA2008-06537 is
also gratefully acknowledged.
7 REFERENCES
Croce P. & Flora A. 2000. Analysis of single fluid jet-grouting.
Geotechnique 50 (6): 739-748.
Croce P. & Modoni G. 2005. Design of Jet Grouting Cut-offs.
Ground Improvement 10 (1): 1-9.
Croce P., Modoni G. & Russo G. 2004. Jet grouting perform-
ance in tunneling. ASCE Geosupport 2004, Orlando, Feb-
ruary 4-7, 2004.
Deu, A., Arroyo, M., Gens, A. & Alonso, E. 2007. Tunnelling
under an active railway line on Barcelona soft soils in Tun-
nels, drivers of change ITA international conference, Ma-
drid November 2007, 858-869. AETOS
ENV 12716. 2001. Execution of special geotechnical works:
jet grouting. European Committee for Standardization.
Guatteri, G., Koshima, A., Lopes, R, Ravaglia, A. & Pieroni,
M.R. 2009. Historical cases and use of horizontal jet grout-
ing solutions with 360◦ distribution and frontal septum to
consolidate very weak and saturated soils. In Ng, Huang &
Liu (eds) 6th International Symposium Geotechnical As-
pects of Underground Construction in Soft Ground –Taylor
& Francis Group, London
Gens, A., Di Mariano, A. & Yubero, T. 2011. EPB tunneling
in deltaic deposits: observations of ground movements. In
7th International Symposium Geotechnical Aspects of Un-
derground Construction in Soft Ground (this conference)
Ishihara, K. & Lee, W.F. 2008. Forensic diagnosis for site-
specific ground conditions in deep excavations of subway
constructions. In Huang &Mayne (eds) Geotechnical and
geophysical site characterization, 31-59
Jean Lutz S.A. 2006. TIGOR, Instrumented rod for the drilling
deviation measurement, Technical sheet
Modoni G., Croce P. & Mongiovì L. 2006. Theoretical model-
ling of jet grouting. Géotechnique, 56 (5): 335-347.
SACYR 2009. High Speed railway line Madrid-Barcelona-
French frontier. Stretch La Torrassa – Sants. SACYR
van Tol, F. 2004. Lessons learned from jet grouting at a tunnel
project in the Hague. In A Dhouib, Magnan et Mestat
(Eds.), ASEP-GI 2004 (321-331). Paris: LCPC